BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Department of Electrical & Computer Engineering - ECPv6.15.18//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:Department of Electrical & Computer Engineering
X-ORIGINAL-URL:https://ece.northeastern.edu
X-WR-CALDESC:Events for Department of Electrical & Computer Engineering
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20210314T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20211107T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20220313T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20221106T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20230312T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20231105T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20240310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20241103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230215T113000
DTEND;TZID=America/New_York:20230215T123000
DTSTAMP:20260423T064101
CREATED:20230210T210554Z
LAST-MODIFIED:20230210T210554Z
UID:6101-1676460600-1676464200@ece.northeastern.edu
SUMMARY:Yiyue Jiang's PhD Proposal Review
DESCRIPTION:“FPGA-based Accelerator of Neural Networks for Digital Predistortion” \nCommittee: \nProf. Miriam Leeser (Advisor) \nProf. John Dooley \nProf. Stefano Basagni \nAbstract: \nPower Amplifiers (PAs) are an essential part of wireless communications. \nAs wireless standards evolve and become more demanding\,  the requirements for PAs change as well.  Specifically\, PAs need to balance linearity and energy efficiency while adhering to 5G wireless standards and beyond. PA behaviors differ based on several criteria\, including the type of PA\, power levels\, and the environment. To overcome the nonlinear behavior of a PA\, a flexible system to achieve digital predistortion (DPD) is required that can rapidly adapt to its environment. \nIn many situations\, traditional methods such as the memory polynomial model cannot adapt to all these factors. Neural networks have been used for some years in RF and microwave engineering. Early work demonstrated the suitability of neural networks to model complicated active device characteristics. Current neural network based DPD systems all do the training offline and are therefore not real-time systems. To reduce the cost to upgrade hardware and to provide more flexibility to different power amplifiers’ linearization needs\, a specific neural network based reconfigurable\, adaptive\, and real-time digital predistortion system is proposed. This system targets Zynq All Programmable System on Chip (SoC) devices which feature an ARM processor and FPGA together with RF frontend on the same chip. The system proposed in this research combines real-time DPD with on-chip training. Furthermore\, most research on FPGA based inference accelerators targets classification problems with probability output. There is no accelerator working on the signal processing problem focusing on sample-by-sample output. Our proposed system is optimized in both algorithm and implementation targeting sample-by-sample processing with high accuracy and real-time efficiency. \n 
URL:https://ece.northeastern.edu/event/yiyue-jiangs-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230203T150000
DTEND;TZID=America/New_York:20230203T170000
DTSTAMP:20260423T064101
CREATED:20230201T200236Z
LAST-MODIFIED:20230201T200236Z
UID:6078-1675436400-1675443600@ece.northeastern.edu
SUMMARY:Kubra Alemdar's PhD Proposal Review
DESCRIPTION:“Overcoming and Engineering Wireless Signals for Communication and Computation” \nAbstract: \nThe phenomenal growth of connected devices\, especially rapid expansion of IoT networks and the increasing demand for wireless services are the main driving forces for the evolution of wireless technologies. However\, the realization of such technologies requires a radical transformation of existing infrastructures to satisfy the needs of changing wireless environments. The main limitation in delivering these systems stems from a huge diversity in their demands and constraints. To address this limitation\, this dissertation shows how wireless signals and their interaction with and within wireless propagation domain can be used as communication or computational tools that enable us to achieve certain novel tasks. Specifically\, we build i) cross-functionality architectures to engineer the wireless channel to a) enable the operation of emerging technologies\, and b) demonstrate a new paradigm for computing with wireless signals\, and ii) intelligently shape the wireless channel to create reliable communication links. \nThis dissertation presents an experimentally validated software-hardware system to deliver three key contributions: We present a physical layer solution for distributed networks that provides over-the-air (OTA) clock synchronization\, called as RFCLOCK\, to overcome the hurdle of implementing fine-grained synchronization for emerging technologies. We first develop the theory for such precision synchronization and second implement it in a custom-design\, which is compatible with commercial-off-the-shelf (COTS) software-defined radios (SDRs). We compare the performance of RFClock with popular wired and GPS-based hardware solutions\, both in terms of clock performance\, as well as impact on distributed beamforming. \nNext\, we propose an RIS-based (reconfigurable reflecting surface) spatio-temporal approach to enhance the link reliability for IoTs where sensors are small-factor designs with single-antenna in rich multipath environment. We demonstrate the design of RIS and how it can effectively perturb the environment\, generating multiple wireless propagation channels and achieving performance of multi-antenna receiver in a Single-Input Single-Output (SISO) link. We compare the performance of the system with multi-antenna receiver in terms of channel hardening and outage probability. \nFinally\, we propose AirFC\, a system harnessing the capability of OTA computation to run inference on a neural network (NN) consisting of a set of fully connected layers (FC) by leveraging multi-antenna systems. We experimentally demonstrate and validate that such computation is accurate enough when compared to its digital counterpart. \nAs part of proposed research ahead\, we will address the challenges of realizing RIS-assisted communication in non-stationary conditions where the wireless channel can abruptly change due to the dynamic environment. We will first demonstrate the conditions in which conventional channel estimation methods cannot be utilized. We will then propose a learning method to create directional beams through reflections from RIS towards target locations without estimating the channel. \nLocation: 632 ISEC \nCommittee: \nProf. Kaushik Chowdhury (Advisor) \nProf. Marvin Onabajo \nProf. Josep Jornet
URL:https://ece.northeastern.edu/event/kubra-alemdars-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20230202T103000
DTEND;TZID=America/New_York:20230202T113000
DTSTAMP:20260423T064101
CREATED:20230126T204948Z
LAST-MODIFIED:20230126T205022Z
UID:6070-1675333800-1675337400@ece.northeastern.edu
SUMMARY:Amani Al-shawabka's PhD Proposal Review
DESCRIPTION:“Channel-and-Adversary-Resilient Radio Fingerprinting through Data-Driven Approaches at Scale” \nCommittee: \nProf. Tommaso Melodia (Advisor)\nProf. Kaushik Chowdhury\nProf. Francesco Restuccia \nAbstract: \nRadio fingerprinting authenticates wireless devices by leveraging tiny hardware-level imperfections inevitably present in the radio circuitry. This way\, devices can be directly identified at the physical layer– thus avoiding energy-expensive upper-layer cryptography that resource-limited embedded devices may not be able to afford. Recent advances have proven that employing deep learning algorithms can achieve fingerprinting accuracy levels that were impossible to achieve by traditional low-dimensional algorithms. Still\, the wireless research community lacks an exhaustive understanding of the challenges associated with developing robust\, reliable\, and channel-resilient radio fingerprinting through deep-learning approaches for practical applications. Key challenges are the non-stationarity of the wireless channel\, and the dynamic effects introduced by the operational environment\, which significantly limit fingerprinting applications by obscuring the hardware impairments associated with the transmitted waveform.\nIn this thesis\, we (i) develop a full-fledged\, systematic investigation to quantify the impact of the wireless channel by providing a first-of-its-kind evaluation on deep-learning-based fingerprinting algorithms\, examining the worst-case scenario (employing devices with identical radio circuitry) and at scale; (ii) develop large-scale open datasets for radio fingerprinting collected in diverse\, rich\, channel conditions and environments\, and using different technologies\, including WiFi and LoRa; (iii) identify conditions where hardware impairments are still detectable; and (iv) design\, implement\, and benchmark new data-driven algorithms to counter the degradation introduced by the wireless channel. Notably\, we propose a generalized\, real-time channel- and adversary-resilient data-driven approach to authenticate wireless devices at scale in practical scenarios. To the best of our knowledge\, our work for the first time improves the fingerprinting accuracy of the worst-case scenario with up to 4x and 6.3x for WiFi and LoRa technologies\, respectively.
URL:https://ece.northeastern.edu/event/amani-al-shawabkas-phd-proposal-review-2/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221209T120000
DTEND;TZID=America/New_York:20221209T133000
DTSTAMP:20260423T064101
CREATED:20221201T022737Z
LAST-MODIFIED:20221201T022840Z
UID:6006-1670587200-1670592600@ece.northeastern.edu
SUMMARY:Alexey Tazin's PhD Dissertation Defense
DESCRIPTION:“Composition of UML Class Diagrams Using Category Theory and External Constraints” \nAbstract:\nIn large software development projects there is always a need for refactoring and optimization of the design. Usually software designs are represented using UML diagrams (e.g class diagrams). A software engineering team may create multiple versions of class diagrams satisfying some external constraints. In some cases\, subdiagrams of the developed diagrams can be selected and combined into one diagram. It is difficult to perform this task manually since manual process is very time consuming\, is prone to human errors\, and is not manageable for large projects. In this dissertation we present an algorithmic support for automating the generation of composed diagrams\, where the composed diagram satisfies a given collection of external constraints and is optimal with respect to a given objective function. The composition of diagrams is based on the colimit operation from category theory. The developed approach was verified experimentally by generating random external constraints (expressed in SPARQL and OWL)\, generating random class diagrams using these external constraints\, generating composed diagrams that satisfy these external constraints\, and computing class diagram metrics for each composed diagram. \nCommittee: \nProf. Mieczyslaw Kokar (Advisor) \nProf. David Kaeli \nDr. Jeff Smith
URL:https://ece.northeastern.edu/event/alexey-tazins-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221209T110000
DTEND;TZID=America/New_York:20221209T130000
DTSTAMP:20260423T064101
CREATED:20221201T023204Z
LAST-MODIFIED:20221201T023204Z
UID:6010-1670583600-1670590800@ece.northeastern.edu
SUMMARY:Bin Sun's PhD Dissertation Defense
DESCRIPTION:“Factorization guided Lightweight Neural Networks for Visual Analysis” \nCommittee: \nProf. Yun Fu (Advisor) \nProf. Ming Shao \nProf. Lili Su \nAbstract: \nDeep learning has become popular in recent years primarily due to powerful computing devices such as GPUs. However\, many applications such as face alignment\, image classification\, and gesture recognition need to be deployed to multimedia devices\, smartphones\, or embedded systems with limited resources. Thus\, there is an urgent need for high-performance but memory-efficient deep learning models. For this\, we design several lightweight deep learning models for different tasks with factorization strategies. \nSpecifically\, we constructed a lightweight face alignment model by proposing a factorization-based deep convolution module named Depthwise Separable Block (DSB) and a light but practical module based on the spatial configuration of the faces. Experiments on four popular datasets verify that Block Mobilenet has better overall performance with less than 1MB storage size.\nBesides the face analysis application\, we also explored a general\, lightweight deep learning module for image classification with low-rank pointwise residual (LRPR) convolution\, called LRPRNet. Essentially\, LRPR aims at using a low-rank approximation to factorize the pointwise convolution while keeping depthwise convolutions as the residual module to rectify the LRPR module. Moreover\, our LRPR is quite general and can be directly applied to many existing network architectures. \nDue to the success of the factorization strategy on image-based data\, we extended factorization on time sequence data for Sign Language Recognition (SLR). We achieved the first rank in the challenge of SLR with the help of our proposed novel Separable Spatial-Temporal Convolution Network (SSTCN)\, which divides a 3D convolution on joint features into several stages \, which help the SSTCN achieve higher accuracy with fewer parameters. \nWe also tried to factorize the features for single image super resolution (SISR). Factorization on features will reduce the feature size in order to reduce the computation costs. However\, the reduction of the spatial size is counter-intuitive for the super resolution task. With our exploration\, we demonstrated a network named Hybrid Pixel-Unshuffled Network (HPUN)\, which factorized the features to achieve the lightweight purpose while keeping high performance. Specifically\, we utilized pixel-unshuffle operation to factorize the input features. After the factorization\, we improved the performance by the grouped convolution\, max-pooling\, and self-residual. The experiments on popular benchmarks showed that the factorization strategy could achieve SOTA performance on SISR.
URL:https://ece.northeastern.edu/event/bin-suns-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221208T140000
DTEND;TZID=America/New_York:20221208T160000
DTSTAMP:20260423T064101
CREATED:20221202T201226Z
LAST-MODIFIED:20221202T201226Z
UID:6014-1670508000-1670515200@ece.northeastern.edu
SUMMARY:Chuangtang Wang's PhD Proposal Review
DESCRIPTION:“All-optical Control of Magnetization in Nanostructures” \nCommittee: \nProf. Yongmin Liu (Advisor) \nProf. Don Heiman \nProf. Nian X. Sun \nAbstract:\nThe switching of magnetization by a femtosecond laser within several picoseconds has recently gained substantial attention\, because it promises next-generation\, energy-efficient\, and high-rate data storage technology. One of the most intriguing demonstrations is the helicity-dependent switching (HD-AOS) of a ferromagnet\, in which the magnetization states can be deterministically written and erased using left- and right-circularly polarized light. However\, the challenge is to realize a single-pulse HD-AOS. Controlling the spin angular momentum transfer from light to magnetic materials in nanostructures is the key to advance this field.\nIn my thesis research work\, I will study the all-optical control of magnetization in different nanostructures\, aiming to better understand the underlying mechanisms of HD-AOD and accelerate the technology development. Firstly\, helicity-driven magnetization dynamics in heavy metal/ferromagnet Au(Pt)/Co bilayer by the optical spin transfer torque (OSTT) is experimentally explored. The wavelength-dependent measurement of OSTT reveals that the quantum efficiency of OSTT strongly depends on the interface electronic structure and pump energy. The Inverse Faraday effect (IFE)\, which is believed to be the driving mechanism of HD-AOS\, is subsequently investigated in an Au thin film. The dependence of IFE on photon energy implies that the orbital angular momentum contribution to IFE is dominated by the excitation of laser pulses. To the best of our knowledge\, it is the first demonstration of this phenomenon. Lastly\, I will discuss our recent results on plasmonics-enhanced all-optical control of magnetization. Light can be tightly confined in plasmonic structures\, which can potentially enable low-energy and high-density magnetic data storage.
URL:https://ece.northeastern.edu/event/chuangtang-wangs-phd-proposal-review/
LOCATION:138 ISEC\, 360 Huntington Ave\, 138 ISEC\, Boston\, MA\, 02115\, United States
GEO:42.3401758;-71.0892797
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=138 ISEC 360 Huntington Ave 138 ISEC Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 138 ISEC:geo:-71.0892797,42.3401758
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221208T110000
DTEND;TZID=America/New_York:20221208T120000
DTSTAMP:20260423T064101
CREATED:20221201T023045Z
LAST-MODIFIED:20221201T023045Z
UID:6008-1670497200-1670500800@ece.northeastern.edu
SUMMARY:Danlin Jia's PhD Dissertation Defense
DESCRIPTION:“Towards Performance and Cost-efficiency for Data-intensive Applications in Distributed Data Processing Systems” \nAbstract: \nData-intensive science (DIS) has experienced a significant boom in the past decade. The emerging technologies of data-intensive services and infrastructures contribute to DIS’s development and raise challenges. An ecosystem has been constructed considering performance\, scalability\, sustainability\, and reliability to provide a high-quality service to DIS applications. The ecosystem consists of services exposed to users for application deployment and infrastructures to support data storage\, transfer\, and management from the system’s perspective. DIS applications share typical features\, such as memory and I/O intensity. Thus\, addressing the bottlenecks triggered by memory-intensive or I/O-intensive workloads in services and infrastructures is essential to improve the performance and cost-efficiency of the whole ecosystem. In this dissertation\, we investigate the characteristics of various DIS applications and design new resource allocation and scheduling schemes for the services and infrastructures in the DIS ecosystem. \nWe first investigate memory optimization in DIS ecosystems. In-memory data analytic frameworks are proposed to cache critical intermediate data in memory instead of in storage drives. Apache Spark is a commonly adopted in-memory data analytic framework with two memory managers\, Static and Unified. However\, the static memory manager lacks flexibility. In contrast\, the unified memory manager puts heavy pressure on the garbage collection of the Java Virtual Machine on which Spark resides. To address these issues\, we propose a new learning-based bidirectional usage-bounded memory allocation scheme to support dynamic memory allocation considering both memory demands and latency introduced by garbage collection. Distributed data-processing workloads in container-based virtualization take advantage of resource sharing\, fast delivery\, and excellent portability of containerization but also suffer from resource competition and performance interference. This inevitably induces performance degradation and significantly long latency\, even worse when over-provisioning. Motivated by this problem\, we design an efficient memory allocation scheme (RITA) for containerized parallel systems to improve data processing latency. RITA monitors applications’ memory usage and cache characteristics and dynamically re-allocates memory resources. \nWe also propose I/O optimizations for DIS applications and infrastructures. Distributed Deep Learning (DDL) accelerates DNN training by distributing training workloads across multiple computation accelerators\, e.g.\, GPUs. Although a surge of research has been devoted to optimizing DDL training\, the impact of data loading on GPU usage and training performance has been relatively under-explored. When multiple DDL applications are deployed\, the lack of a practical and efficient technique for data-loader allocation incurs GPU idleness and degrades the training throughput. In this dissertation\, we thus investigate the impact of data-loading on the global training throughput and design a resource allocator that uses the data-loading rate as a knob to reduce the GPU idleness. Finally\, designs and optimizations on disaggregated storage systems supported by cutting-edge storage and network techniques emerge dramatically. Disaggregated storage systems can scale resources independently and provide high-quality services for hyper-scale architectures. The traditional congestion control mechanism relieves congestion by limiting the data-sending rate of senders. However\, such a design scarifies the storage drive’s performance as data are generated but stalled on storage host nodes if network congestion happens. To solve this issue\, we design a storage-side rate control mechanism to mitigate network congestion while avoiding sacrificing I/O performance. \nCommittee: \nProf. Ningfang Mi (Advisor) \nProf. Xue Lin \nProf. David Kaeli
URL:https://ece.northeastern.edu/event/danlin-jias-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221206T160000
DTEND;TZID=America/New_York:20221206T173000
DTSTAMP:20260423T064101
CREATED:20221206T020005Z
LAST-MODIFIED:20221206T020005Z
UID:6017-1670342400-1670347800@ece.northeastern.edu
SUMMARY:Md Navid Akbar's PhD Dissertation Defense
DESCRIPTION:“Inference from Brain Imaging: Incorporating Domain Knowledge and Latent Space Modeling” \nAbstract:\n\nBrain imaging can probe the anatomy (structural) of our brain\, or its function (functional). A particular imaging modality (unimodal) generally provides only a particular insight into human health. Transcranial magnetic stimulation (TMS)\, though still in its infancy as a brain imaging modality\, is such a functional\, unimodal technique. TMS helps model human motor-cortical mapping\, using corresponding muscle activity captured by surface electromyography (EMG)\, but it necessitates a reliable data-driven model. Earlier works have modeled the causal direction only (from cortical representation to muscles)\, or the inverse direction (from muscles to cortical representation)\, with simple statistical regression. We modeled this motor-cortical mapping bi-directionally in this dissertation\, using deep learning. We first modeled TMS-induced 3D electric field (E-field) in a brain to causal multi-muscle activation picked up by EMG\, in a regression task using a convolutional neural network (CNN) autoencoder. By fusing neuroscience domain knowledge (e.g.\, an empirical neural response profile)\, we reduced 14% squared error\, compared to the baseline model that did not contain this. We then designed our novel inverse imaging CNN model\, to reconstruct physiologically meaningful E-field distributions (in the image domain) from a given set of muscle activations (in the sensor domain). By adopting variational inference in the CNN model\, to learn the underlying latent space better\, we were able to reduce 13% in squared error over our purely CNN baseline. \nDiagnosis with brain imaging is often incomplete with a unimodal technique\, and having multiple sources (multimodal) may be advantageous. Successful multimodal fusion can provide more holistic information\, compared to its constituents. One relevant example is the classification of late post-traumatic seizure (LPTS). Previous works in this space have tackled LPTS classification with either unimodal functional imaging\, or non-machine learning (ML) structural modeling. In this dissertation\, we first undertook the ML classification of binary LPTS: with unimodal\, structural brain imaging\, namely diffusion magnetic resonance imaging (dMRI). By incorporating interpretable domain knowledge (post-traumatic lesion volume compensation)\, we improved 7% in the mean area under the curve (AUC) over the standard technique in literature. Finally\, we classified LPTS for a larger sample of subjects\, utilizing multimodal imaging\, including functional MRI (fMRI) and electroencephalography (EEG). Following unsupervised imputation for any missing modality within the subjects\, we introduced our novel multimodal fusion algorithm\, which attempts to leverage the underlying structure of the multivariate information. We found that our proposed algorithm improved by 7% in AUC performance\, over a naive Bayesian estimator that can handle missing data intrinsically.\nCollectively\, the work presented here demonstrated that incorporating domain knowledge in the modeling pipeline successfully improved inference. Similar improvements were also observed by learning and leveraging the possible underlying latent structure of the given information\, and adapting the models accordingly. \n\n\n\nCommittee:\n\nProf. Deniz Erdogmus (Advisor) \nProf. Mathew Yarossi (Co-advisor)\nProf. Dominique Duncan\nProf. Sarah Ostadabbas
URL:https://ece.northeastern.edu/event/md-navid-akbars-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221129T140000
DTEND;TZID=America/New_York:20221129T153000
DTSTAMP:20260423T064101
CREATED:20221122T012209Z
LAST-MODIFIED:20221122T012209Z
UID:5975-1669730400-1669735800@ece.northeastern.edu
SUMMARY:Prof. Hui Guan -  "Towards accurate and efficient edge computing via multi-task learning "
DESCRIPTION:“Towards accurate and efficient edge computing via multi-task learning ” \n\nAbstract: \n\n\nAI-powered applications increasingly adopt Deep Neural Networks (DNNs) for solving many prediction tasks\, leading to more than one DNNs running on resource-constrained devices. Supporting many models simultaneously on a device is challenging due to the linearly increased computation\, energy\, and storage costs. An effective approach to address the problem is multi-task learning (MTL) where a set of tasks are learned jointly to allow some parameter sharing among tasks. MTL creates multi-task models based on common DNN architectures and has shown significantly reduced inference costs and improved generalization performance in many machine learning applications. In this talk\, we will introduce our recent efforts on leveraging MTL to improve accuracy and efficiency for edge computing. The talk will introduce multi-task architecture design systems that can automatically identify resource-efficient multi-task models with low inference costs and high task accuracy. \n\n\nBio:\n \n\n\n\nHui Guan is an Assistant Professor in the College of Information and Computer Sciences (CICS) at the University of Massachusetts Amherst\, the flagship campus of the UMass system. She received her Ph.D. in Electrical Engineering from North Carolina State University in 2020. Her research lies in the intersection between machine learning and systems\, with an emphasis on improving the speed\, scalability\, and reliability of machine learning through innovations in algorithms and programming systems. Her current research focuses on both algorithm and system optimizations of deep multi-task learning and graph machine learning.
URL:https://ece.northeastern.edu/event/prof-hui-guan-towards-accurate-and-efficient-edge-computing-via-multi-task-learning/
LOCATION:442 Dana\, 360 Huntington Ave\, 442 DA\, Boston\, MA\, 02115\, United States
GEO:42.3387508;-71.0923044
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=442 Dana 360 Huntington Ave 442 DA Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 442 DA:geo:-71.0923044,42.3387508
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221129T100000
DTEND;TZID=America/New_York:20221129T130000
DTSTAMP:20260423T064101
CREATED:20221104T010151Z
LAST-MODIFIED:20221104T010151Z
UID:5952-1669716000-1669726800@ece.northeastern.edu
SUMMARY:Research Presentations On Bendable Electronics and Sustainable Technologies (BEST)
DESCRIPTION:Professor Ravinder Dahiya will be joining Northeastern’s ECE Department on January 2023. Please join us for an interactive mini-symposium featuring projects from the BEST Lab directed by Professor Dahiya. \n  \nThe presenters are: \nDr. Dhayalan Shakthivel\, Research Associate\, Inorganic Nanowires for Flexible and Large Area Electronics \nDr. Gaurav Khandelwal\, Post-doc\, Functional Materials based Triboelectric Nanogenerators for Selfpowered Sensors and Systems \nDr. Fengyuan Liu\, Post-doc\, “Hebbian-like” learning in electronic skin \nDr. Abhishek S. Dahiya\, Research Associate\, Towards energy autonomous electronic skin using sustainable technologies \nAyoub Zumeit\, PhD candidate\, Inorganic nanostructures-based high-performance flexible electronics \nAdamos Christou\, PhD candidate\, Novel Technologies for High-Performance Printed Electronics \nRadu Chirila\, PhD candidate\, Electronic Skin and Holographic Systems for Socially Intelligent Robots \nJoão Neto\, PhD candidate\, Hardware building for neuromorphic electronic skin \nLuca De Pamphilis\, PhD candidate\, Nanowire-based electronic layers for flexible neuromorphic devices \nMake sure to RSVP & specify inperson or virtual attendance. See you soon!
URL:https://ece.northeastern.edu/event/research-presentations-on-bendable-electronics-and-sustainable-technologies-best/
LOCATION:442 Dana\, 360 Huntington Ave\, 442 DA\, Boston\, MA\, 02115\, United States
GEO:42.3387508;-71.0923044
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=442 Dana 360 Huntington Ave 442 DA Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave\, 442 DA:geo:-71.0923044,42.3387508
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221128T120000
DTEND;TZID=America/New_York:20221128T140000
DTSTAMP:20260423T064101
CREATED:20221121T212045Z
LAST-MODIFIED:20221121T212045Z
UID:5973-1669636800-1669644000@ece.northeastern.edu
SUMMARY:Xuanyi Zhao's PhD Proposal Review
DESCRIPTION:“AlN/AlScN based Micro Acoustic Metamaterials for Radio Frequency Applications of the Next Generations” \nAbstract: \nIn the last two decades‚ micro-acoustic resonators (μARs) have played a key role in integrated 1G-to-4G radios‚ providing the technological means to achieve compact radio frequency (RF) filters with low loss and moderate fractional bandwidths (BW<4%). More specifically‚ Aluminum Nitride (AlN) based filters have populated the front-end of most commercial mobile transceivers due to the good dielectric‚ piezoelectric and thermal properties exhibited by AlN thin-films and because their fabrication process is compatible with the one used for any Complementary Metal Oxide Semiconductor (CMOS) integrated circuits (ICs). Nevertheless‚ the rapid growth of 5G and the abrupt technological leap expected with the development of sixth-generation (6G) communication systems are expected to severely complicate the design of future radio front-ends by demanding Super-High-Frequency (SHF) filtering components with much larger fractional bandwidths than achievable today. In the meantime\, as more acoustic filters replying on bulk waves which requests the devices to be physically-suspended to operate\, thermal related nonlinearity has been a challenge which requests new designs to enhance the thermal linearity thus power handling for these acoustic components. Even more‚ the recent invention of on-chip nonreciprocal components‚ like the circulators and isolators recently built in slightly different CMOS technologies‚ has provided concrete means to double the spectral efficiency of current radios by enabling the adoption of full-duplex communication schemes. Nevertheless‚ for such schemes to be really usable in wireless systems‚ self-interference cancellation networks including wideband‚ low-loss and large group delay lines are needed. Yet‚ the current on-chip delay lines that are also manufacturable through CMOS processes‚ which rely on the piezoelectric excitation of Surface Acoustic Waves (SAWs) or Lamb Waves in piezoelectric thin films‚ have their bandwidth and insertion-loss severely limited by the relatively low electromechanical coupling coefficient exhibited by their input and output transducers. As a results‚ these components are hardly usable to form any desired self-interference cancelation networks. In order to overcome these challenges‚ only recently‚ new classes of microacoustic resonators and delay lines exploiting the high piezoelectric coefficient of Aluminum Scandium Nitride (AlScN) thin films and the exotic dispersive features of acoustic metamaterials (AMs) have been emerging. These devices rely on forests of locally resonant piezoelectric rods to generate unique modal distributions‚ as well as unconventional wave propagation features that cannot be found in conventional SAW and Lamb wave counterparts. In this presentation‚ the design‚ fabrication and performance of the first microacoustic metamaterials (μAMs) based resonators and delay lines will be showcased. Moreover\, AMs based reflectors are invented and demonstrated providing new improving the linearity and power handling of the AlScN μARs. In addition to reviewing the current status of our work\, we will propose several further explorations of using our AlN/AlScN based AMs in RF applications of the next generations. \nCommittee: \nProf. Cristian Cassella (advisor) \nProf. Matteo Rinaldi \nDr. Jeronimo Segovia-Fernandez
URL:https://ece.northeastern.edu/event/xuanyi-zhaos-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221122T110000
DTEND;TZID=America/New_York:20221122T120000
DTSTAMP:20260423T064101
CREATED:20221103T213322Z
LAST-MODIFIED:20221103T213322Z
UID:5942-1669114800-1669118400@ece.northeastern.edu
SUMMARY:Mahshid Asri's Proposal Review
DESCRIPTION:“Development of Anomaly Detection and Characterization Algorithms Using Wideband Radar Image Processing for Security Applications” \nAbstract:\nDetection and characterization of suspicious body-worn objects is necessary for safe and effective personnel screening. In airports\, developing a precise system that can distinguish threats and explosives from objects like money belt can reduce the pat-down significantly while maintaining effective security.\nThis work proposes two main algorithms which are developed for different millimeter-wave radar systems. The first project is a material characterization algorithm designed for a 30 GHz wideband multi bi-static radar system used for passenger screening in airports. The proposed algorithm can automatically distinguish lossless materials from lossy ones and calculate their thickness and permittivities. Starting from the radar reconstructed image showing a cross-section of the body\, we extract the nominal body contour using Fourier series\, separate body and object responses\, categorize the object as lossy or lossless based on the depression and protrusion of the body contour\, and finally predict possible values for the object’s permittivity and thickness. Our resulting classification is good\, implying fewer nuisance alarms at check points. The second project is a metal detection algorithm designed to monitor pedestrians walking along a sidewalk for large\, concealed metallic objects. Finite Difference Frequency Domain and SAR algorithms are used to simulate the images produced by this 6 GHz wideband radar system. \nCommittee: \nProf. Carey Rappaport (Advisor) \nProf. Charles DiMarzio \nProf. Edwin Marengo
URL:https://ece.northeastern.edu/event/mahshid-asris-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221118T110000
DTEND;TZID=America/New_York:20221118T120000
DTSTAMP:20260423T064101
CREATED:20221115T232757Z
LAST-MODIFIED:20221115T232757Z
UID:5962-1668769200-1668772800@ece.northeastern.edu
SUMMARY:PhD Dissertation Defense Shivang Aggarwal
DESCRIPTION:Location: ISEC 332 \n“Towards Reliable\, High Capacity mmWave Wireless LANs for Mobile Devices” \nAbstract: \nThe IEEE 802.11ad standard\, with its 14 GHz of unlicensed spectrum around 60 GHz\, is touted as one of the key technologies for building the next generation of WLANs that will enable high throughput demanding mobile applications. However\, there have been serious concerns regarding the susceptibility of mmWave links to mobility and blockage as well as smartphone energy consumption at Gigabit scale data rates. \nIn this dissertation\, first\, through extensive measurement campaigns with commercial off-the-shelf (COTS) devices as well as a highly configurable software-defined radio (SDR) based testbed\, we characterize the performance and energy efficiency of mobile devices operating in 60 GHz WLANs and identify problems that prevent wide adoption of the mmWave technology in such devices. Then\, using the insights from these measurement campaigns\, we design solutions to tackle these problems and prototype them for real-world evaluation.\nThis dissertation makes the following contributions:\n(i) We extensively study the performance and power consumption of 802.11ad on commercial smartphones. We focus on the specific aspects affected by unique smartphone features\, e.g.\, antenna placement or user mobility patterns\, and compare the performance against that achieved by 802.11ad laptops in previous studies. We also compare 802.11ad against its main competitors 802.11ac and 802.11ax. Overall\, our results show that 802.11ad is better able to address the needs of emerging bandwidth-intensive applications in smartphones than its 5 GHz counterparts. At the same time\, we identify several key research directions towards realizing its full potential.\n(ii) We extensively study the two main link adaptation mechanisms in 802.11ad\, rate adaptation (RA) and beamforming. We undertake a large measurement campaign using an SDR-based testbed giving us complete access to the PHY and MAC layers. We look at the two link adaptation mechanisms separately and study the effectiveness of a few RA and beamforming heuristics. Further\, look at the interaction between the two link adaptation mechanisms\, specifically\, which mechanism should be triggered when and in what order. We design a practical\, standard-compliant link adaptation framework that leverages ML and PHY layer information to determine when to trigger link adaptation and which adaptation mechanism to use.\n(iii) To address the issues with mmWave link reliability\, we explore the use of multiple frequency bands to couple the performance of 802.11ad with the reliability of legacy WiFi. To accomplish this\, we develop a Multipath TCP (MPTCP) scheduler to efficiently use both interfaces simultaneously in order to achieve a higher overall throughput as well as seamlessly switch to a single interface when the other one fails. Further\, we port MPTCP to a dual-band (5 GHz/60 GHz) smartphone\, evaluate its power consumption\, and provide recommendations towards the design of an energy-aware MPTCP scheduler.\n(iv) To enable high user QoE\, and maintain that in the face of ever-changing network conditions\, applications such as virtual reality (VR) and video streaming perform quality adaptation. A key component of quality adaptation is throughput prediction. Thus\, we extensively study the predictability of the network throughput of an 802.11ad WLAN in downloading data to an 802.11ad- enabled mobile device under varying mobility patterns and orientations of the mobile device.\n(v) With a dramatic increase in throughput requirements of applications and AP-user density in the near future\, multi-user multi-stream communication in the 60 GHz band is required. To this end\, the IEEE 802.11ay standard\, successor to the current 802.11ad standard\, includes support for simultaneous transmission over multiple data streams. Using an SDR-based testbed\, we extensively study the performance of SU- and MU-MIMO in 60 GHz WLANs in multiple environments\, analyze the performance in each environment\, identify the factors that affect it\, and compare it against the performance of SISO. Finally\, we propose two heuristics that perform both beam and user selection with low overhead while outperforming previously proposed approaches \nCommittee:\nProf. Dimitrios Koutsonikolas (Advisor)\nProf. Kaushik Chowdhury\nProf. Tommaso Melodia
URL:https://ece.northeastern.edu/event/phd-dissertation-defense-shivang-aggarwal/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221115T140000
DTEND;TZID=America/New_York:20221115T150000
DTSTAMP:20260423T064101
CREATED:20221103T184043Z
LAST-MODIFIED:20221115T204305Z
UID:5920-1668520800-1668524400@ece.northeastern.edu
SUMMARY:Raana Sabri Khiavi's PhD Proposal Review
DESCRIPTION:“Theory and design of spatiotemporally-modulated metasurfaces for comprehensive control of light” \nAbstract: \nPhotonic metasurfaces are key platforms for manipulating almost all properties of light such as amplitude\, phase\, polarization\, wave vector\, pulse shape\, and orbital angular momentum in a sub-wavelength dimension. They are capable of providing unprecedented modulation of wavefront through imparting spatial or temporal variation on the incoming wave. Recently\, considerable efforts have been devoted to design active metasurfaces that enable real-time tuning and post-fabrication control of the optical response. Toward achieving this goal\, electro-optically tunable materials such as doped semiconductors\, multiple-quantum-wells (MQWs)\, and atomically thin sheets are incorporated into the building blocks of the geometrically-fixed metasurfaces. Despite the significant progress in this field\, there has been several limitations imparted to the optical response of such so-called quasi-static metasurfaces. Remarkably\, the strong resonant dispersion in such metasurfaces leads to narrow spectral and angular bandwidths. In addition\, the co-varying amplitude and phase response as well as the limited phase modulation give rise to undesired artefacts manifested on their output profiles. The slow response time to the external stimuli is another drawback that restricts the performance of the metasurfaces. Introducing time into the external stimulus of the metasurfaces\, as an additional degree of freedom\, offers a way out to surmount the obstacles facing the quasi-static metasurfaces. Modulation in time enables myriad of exotic space-time scattering phenomena\, where possibility to break the reciprocity and generation/manipulation of the sideband scattered signals offer the most appealing functionalities. The objective of this work is to investigate the less explored mechanisms for yielding reconfigurable plasmonic metasurfaces in both space and time. Several realizations of quasi-static and time-modulated devices integrated with the electro-optical materials such as indium-tin-oxide (ITO) with the potential wide phase modulation is presented. It has been shown that time-modulated metasurfaces are superior to their quasi-static counterparts in terms of providing access to the dispersionless modulation-induced phase shift spanning over 2π as well as the constant amplitude at the sidebands. Novel and unique applications of space-time photonic metasurfaces by spatiotemporal manipulation of light for all-angle\, broadband beam steering\, suppressing the undesired sidelobes\, high speed continuous beam scanning\, and dispersionless dynamic wavefront engineering are studied. \nCommittee: \nProf. Hossein Mosallaei (Advisor) \nProf. Charles DiMarzio \nProf. Siddhartha Ghosh
URL:https://ece.northeastern.edu/event/raana-sabri-khiavis-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221108T153000
DTEND;TZID=America/New_York:20221108T163000
DTSTAMP:20260423T064101
CREATED:20221103T213536Z
LAST-MODIFIED:20221103T213536Z
UID:5948-1667921400-1667925000@ece.northeastern.edu
SUMMARY:Giuseppe Michetti's PhD Dissertation Defense
DESCRIPTION:“RF Front-End Components based on Linear-Time-Variant Modulation of Piezoelectric MEMS Resonators” \nAbstract: \nThroughout the last decade\, radio frequency (RF) components for over-the-air communication and sensing have been subject to sustained market pressure to adapt to the novel trends such as spectrum sharing\, programmability\, and low-power operation. When these features are required in chip-scale RF hardware\, innovative solutions are necessary as conventional materials and techniques become bottlenecks for next-generation radios. In this work\, we explore advanced wave manipulation circuital techniques such as Linear-Time-Variant (LTV) networks in conjunction with high-performance RF passives based on Micro-Electro-Mechanical Systems (MEMS) to address some of these challenges. Leveraging the unique spectral characteristic of RF MEMS resonators\, we show some components based on LTV concepts\, for novel RF systems with advanced spectral efficiency and real-time reconfigurability. \nUsing AlN and ScAlN thin film MEMS resonators as building blocks\, we propose a design technique for MEMS-based LTV Circulators and Self Interference Cancelers\, enabling chip-scaled RF full-duplex systems to enable efficient use of the RF spectrum with up to 47.5 dB cancellation in an 8 % bandwidth (BW) at 450 MHz. We introduce and validate experimentally MEMS-based LTV BW-tunable filters with high linearity (>30 dBm)\, and 5:1 BW tunability\, designed for several bands from 100 MHz to 2.7 GHz for emerging paradigms such as software-defined-radios and cooperative networks. We also introduce MEMS-based near-zero energy RF front-end for the Internet-of-Things (IoT)\, implementing RF energy harvesting to power up a resonant Wake-Up Receiver circuit\, with an experimental demonstration at (800 MHz) for deployment in remote sensor networks and emerging IoT wearable applications. \nAlong with the experimental validation of the proposed components\, analytical and numerical tools are also discussed for future development and research. \nCommittee: \nProf. Matteo Rinaldi (Advisor) \nProf. Cristian Cassella \nProf. Andrea Alù
URL:https://ece.northeastern.edu/event/giuseppe-michettis-phd-dissertation-defense/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221107T123000
DTEND;TZID=America/New_York:20221107T140000
DTSTAMP:20260423T064101
CREATED:20221103T191749Z
LAST-MODIFIED:20221103T191749Z
UID:5938-1667824200-1667829600@ece.northeastern.edu
SUMMARY:Tianyu Dai's PhD Dissertation Defense
DESCRIPTION:“Robust Data-Driven Control” \nAbstract: \nDuring the last two decades\, data-driven control (DDC) has attracted growing attention in the control community. Unlike model-based control (MBC)\, which first uses the collected data to identify the system\, then designs the controller according to the certainty equivalence principle\, DDC skips the system identification (SYSID) step and leads to a control law directly from data. One crucial feature of DDC is that some fundamental limitations of MBC\, such as uncertainty versus robustness\, inevitable modeling error\, and possible expensive cost of SYSID\, are avoided in the DDC framework. These benefits enable the researcher to design controllers with better performance and accuracy. \nRobust data-driven control (RDDC) as a branch of DDC has developed rapidly in recent years\, focusing on the data-driven controller design for the state space model. It aims to solve the following problem: given a single trajectory of noisy data and a few priors of the model structure\, how to design a robust state feedback controller to stabilize the system with unknown dynamics\, and in addition\, to meet some performance criteria. By robust\, we mean the learned controller can stabilize all possible systems residing in the set compatible with the noisy data. \nThis dissertation aims to summarize our contributions to the RDDC field. We focus on the L_infinity bounded noise\, and the main idea hinges on duality theory to establish the connection between two sets\, one compatible with the noisy data and the second satisfying some design properties such as stability or optimality. Our main results show that for all possible systems compatible with the data\, the data-driven control law can be obtained by solving a convex optimization problem. In the dissertation\, we propose RDDC algorithms for linear\, switched\, and nonlinear systems with process noise\, extend results for error-in-variables (a more general case)\, and discuss a worst-case optimal estimation of the trajectory of a switched linear system. \nCommittee: \nProf. Mario Sznaier (Advisor) \nProf. Octavia Camps\nProf. Bahram Shafai \nProf. Eduardo Sontag
URL:https://ece.northeastern.edu/event/tianyu-dais-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221102T140000
DTEND;TZID=America/New_York:20221102T150000
DTSTAMP:20260423T064101
CREATED:20221103T213638Z
LAST-MODIFIED:20221103T213638Z
UID:5950-1667397600-1667401200@ece.northeastern.edu
SUMMARY:Kai Huang's PhD Dissertation Defense
DESCRIPTION:“Partitioning Data Across Multiple\, Network Connected FPGAs with High Bandwidth Memory to Accelerate Non-streaming Applications” \nAbstract:\nField Programmable Gate Arrays (FPGAs) are increasingly used in cloud computing to increase the run time of various applications. Flexibility\, efficiency and lower power enable FPGAs to be important components in modern data centers. Applications such as Secure Function Evaluation (SFE)\, graph processing\, and machine learning are increasingly mapped to FPGA-based adaptable cloud computing platforms. However\, due to resource limitations\, it is difficult to map applications to only one FPGA. Applications with a streaming data processing pattern can be mapped to a multiple-FPGA platform where the FPGAs are connected in a 1-D or ring topology\, thus communications overhead can be pipelined with computations. The communication\, merely passing data from boards to boards\, will not significantly affect the system performance if the bandwidth is sufficient. In a more general processing pattern involving non-streaming applications\, each FPGA is responsible for only a portion of the computation and the FPGAs must keep exchanging data during the run time of the application. The communication cost can be the bottleneck of such a system. The challenge is how to map and parallelize these applications to a multi-FPGA cloud computing platform in such a way that communication is minimized and speedup is maximized.\nIn this research\, we build a framework to map garbled circuit applications\, an implementation of SFE\, to a cloud computing platform that has FPGA cards attached to computing nodes. The FPGAs on the node are able to communicate directly through the network. The framework consists of two parts: hardware design and software preprocessing. The hardware design integrates with the Xilinx UDP network stack enabling the capability to exchange data through the network and thus bypassing the processor and its software stack. The framework also takes advantage of High Bandwidth Memory (HBM) for high off-chip memory throughput. The levels of memory hierarchy available on the FPGA are used for caching both local data and incoming and outgoing network data. Preprocessing will generate the reordered batches of each layer needed for processing\, efficient memory allocation and final memory layout. We also applied an effective partitioning algorithm to schedule executions to different FPGAs to minimize the communication between FPGAs. By generating different size of problems from the EMP-toolkit\, we can demonstrate that this hardware-software co-design framework achieves nearly optimal two times speedup on a two-FPGA setup compared to a one-FPGA implementation. We explore extremely large examples that cannot be mapped to one-FPGA\, proving that it is achievable to map large examples of billions of operations to this distributed heterogeneous system. \nCommittee: \nProf. Miriam Leeser(advisor) \nProf. Stratis Ioannidis(co-advisor) \nProf. Mieczyslaw Kokar
URL:https://ece.northeastern.edu/event/kai-huangs-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221102T120000
DTEND;TZID=America/New_York:20221102T130000
DTSTAMP:20260423T064101
CREATED:20221103T213443Z
LAST-MODIFIED:20221103T213443Z
UID:5946-1667390400-1667394000@ece.northeastern.edu
SUMMARY:Yuexi Zhang's PhD Proposal Review
DESCRIPTION:“Human Body and Activity Analysis” \nAbstract: \nHuman-related applications such as person detection\, human pose estimations and human activity recognition\, that always draw a lot of attentions in computer vision community. In this proposal\, we discuss several related topics that we are interested in\, and demonstrate how we improve the existing methods. The first problem we consider is video-based human pose estimation. For most general approaches\, researchers focus on collecting human poses from each frame independently and then associate them based on matching or tracking methods. However\, such the pipeline usually relies on complex computations and also consumes running time. To overcome such shortages\, we propose a light weighted network with the unsupervised training strategy\, that aims to reduce running time but remaining the performance. The next problem we explore is about cross-view action recognition (CVAR). The goal of CVAR is to recognize a human action when observed from a previously unseen viewpoint. This is important for some applications such as surveillance systems where is not practical or feasible to collect large amounts of training data when adding a new camera. In this case\, it requires methods that are able to generate reliable view-invariant information trained with given viewpoints and recognize the action from an unseen viewpoint. In general\, most approaches rely on 3D data\, but using 2D data is still under-discovered. Besides\, the performance of those approaches using only 2D data is far worse than 3D approaches. Therefore\, we propose a simple yet efficient CVAR framework that takes 2D data as input and close the performance gap between 3D and 2D input. The last problem we investigate is online action detection and we are interested in detecting action start at current stage. Online action start detection problem is to detect an action startpoint as soon as it occurs with its action category in untrimmed\, streaming videos\, and it has potential applications such as early alert generation in surveillance systems. The typical approaches usually heavily rely on frame-level annotations and also they are limited to pre-defined action categories. Therefore\, we propose a novel yet simple design\, 3D MLP-mxier based architecture that aims to detect the taxonomy-free action start without using frame-level annotations. \n  \nCommittee: \nDr. Octavia Camps(Advisor) \nDr. Mario Sznaier \nDr. Sarah Ostadabbas
URL:https://ece.northeastern.edu/event/yuexi-zhangs-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221028T130000
DTEND;TZID=America/New_York:20221028T140000
DTSTAMP:20260423T064101
CREATED:20221103T213401Z
LAST-MODIFIED:20221103T213401Z
UID:5944-1666962000-1666965600@ece.northeastern.edu
SUMMARY:Guillem Reus Muns' PhD Proposal Review
DESCRIPTION:Location: ISEC 332 \n“AI for communications and sensing in RF environments” \nAbstract: \nThe recent growth of Internet of Things (IoT)\, as well as other new revolutionary applications utilizing wireless spectrum are leading the way towards realization of next generation wireless systems that jointly utilize communications and sensing. However\, such systems offer many degrees of freedom\, and optimizing them for a specific task is difficult to accomplish with deterministic and classical approaches. For this reason\, data-driven and AI-based methods have been pursued actively by the research community\, as they are able to find solutions that often come close to or exceed the performance of the deterministic counterparts with a fractional execution complexity. This thesis presents\, through real systems and with experimental validation\, our progressive efforts in three broad areas\, where AI enables the operation of aerial and terrestrial systems that combine sensing and communications. This dissertation explores the following key use cases with distinct contributions made in each: \ni) Sensing-aided communications for air and ground systems. First\, we present a UAV communication method that defines constellation points in space that map to transmitter frequency bands and are detected at the Base Station using millimeter wave sensors. Second\, we explore alternative vehicle-to-infrastructure mmWave beamforming methods\, leveraging a) vehicle position and velocity estimation using in-band standard compliant 802.11ad radar and b) camera images and GPS location information.\nii) Signal classification using communication signals\, where we propose a) a UAV classification method using uniquely UAV-transmitted signals and b) an RF fingerprinting technique that improves class separation by combining triplet loss with regular classification techniques.\niii) ‘AirFC’\, an over-the-air computation method that implements fully connected neural networks inference leveraging multi-antenna systems. \nFinally\, the proposed work will address challenges in the CBRS band\, where a tiered structure is implemented to access the spectrum. Hence\, continuous sensing is needed to make sure that radar (tier 1) is not interfered by cellular systems (tier 2). Here\, we propose reusing the already existing cellular infrastructure to act as a radar detector\, which enhances their functionality to go beyond that of regular wireless communications. \nCommittee: \nProf. Kaushik Chowdhury (Advisor) \nProf. Hanumant Singh \nProf. Stratis Ioannidis
URL:https://ece.northeastern.edu/event/guillem-reus-muns-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221024T110000
DTEND;TZID=America/New_York:20221024T120000
DTSTAMP:20260423T064101
CREATED:20221103T213221Z
LAST-MODIFIED:20221103T213221Z
UID:5940-1666609200-1666612800@ece.northeastern.edu
SUMMARY:Yixuan He's PhD Proposal Review
DESCRIPTION:Committee: \nProf. Yong-Bin Kim. Advisor \nProf. Marvin Onabajo \nProf. Lombardi Fabrizio \n  \nAbstract: \nIn order to match the needs of powerful neural networks and meet the hard constraints from hardware\, binary neural networks are treated as hardware-friendly deep learning algorithms due to the fact that it can achieve similar inference accuracy with fewer computing resources comparing to traditional convolutional neural networks. As for its VLSI implementations\, the computing-in-memory (CIM) technology has been proved to solve the memory-wall bottleneck problem shown in traditional von Neumann machine and can be a perfect choice to implement neural networks with binary data. Therefore\, this work proposes a novel time-domain computing-in-memory core that implements XNOR-and-accumulate of binary neural networks with all-digital elements. This new technique uses 8T-SRAM cells to perform XNOR operations inside memory array and accumulates the related XNOR output values in time-domain with specialized racing structures and delay lines. The circuit is built and simulated in Cadence using Samsung 65nm CMOS technology with 1V power supply. The results show correct functionality\, 2730 GOPS throughput and 431 TOPS/W power efficiency. With further exploration\, the time-domain computation can be a new candidate in the field of in-memory-computing for deep learning applications since it has its own superiorities in terms of throughput\, power efficiency in comparison to other mixed-signal or traditional digital methods.
URL:https://ece.northeastern.edu/event/yixuan-hes-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221020T120000
DTEND;TZID=America/New_York:20221020T130000
DTSTAMP:20260423T064101
CREATED:20221103T191700Z
LAST-MODIFIED:20221103T191700Z
UID:5936-1666267200-1666270800@ece.northeastern.edu
SUMMARY:Neset Unver Akmandor's PhD Proposal Review
DESCRIPTION:“Improving Computational Efficiency of Motion Planning Algorithms for Mobile and Time-Dependent Robotic Tasks in Dynamic Environments” \nAbstract: \nRobots will become a part of our lives at home as personal assistants. Although their current functionality is highly restricted to specific tasks and environments\, their practicality encourages robotics engineers for further advancement. Especially\, mobile robots with manipulation capabilities have a huge potential to support humans in physically demanding workplaces\, such as warehouses and hospitals. Considering the complexity of the human level tasks and the dynamic settings\, the state-of-the-art robot motion planning methods need to be improved in terms of their computational efficiency. To contribute on closing the gap\, this proposal presents three novelties whose applications focus on mobile robots in dynamic environments. First\, we introduce a reactive navigation framework in 3D workspaces. The proposed approach does not rely on the global map information and achieves fast navigation by employing motion primitives and their heuristic evaluations on the-fly. Second\, we present a Deep Reinforcement Learning based navigation approach in which we define the occupancy observations as heuristic evaluations of motion primitives\, rather than using raw sensor data. It utilizes occupancy observations in different data structures to analyze their effects on both training process and navigation performance. We train and test our methodology on two different robots within challenging physics-based simulation environments including static and dynamic obstacles. Finally\, we propose a computationally efficient framework for trajectory planning for robots with high degrees-of freedom while adapting its system model\, constraints and time-dependent target state using the latest information from the dynamic environment. \n  \nCommittee: \nDr. Taskin Padir (Advisor)Dr. Pau ClosasDr. Michael EverettDr. Erdal Kayacan
URL:https://ece.northeastern.edu/event/neset-unver-akmandors-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221017T130000
DTEND;TZID=America/New_York:20221017T140000
DTSTAMP:20260423T064101
CREATED:20221103T191510Z
LAST-MODIFIED:20221103T191510Z
UID:5934-1666011600-1666015200@ece.northeastern.edu
SUMMARY:Sila Deniz Calisgan's PhD Dissertation Defense
DESCRIPTION:“ADVANCEMENTS ON ZERO STANDBY POWER MEMS SENSORS” \nAbstract: \nDue to the fast development of the internet of things\, and unattended wireless sensor networks\, the number of connected devices worldwide is expected to increase exponentially in the future. In order to maintain such large networks of physical and virtual objects\, there is a need for sensors\, actuators and devices with dimensions and power consumption that are orders of magnitude smaller than the state-of-the-art. Currently no existing technology could enable the implementation of large-scale wireless sensor networks in remote locations due to the prohibitive cost associated with installation and maintenance. The fundamental technical challenge lies in the continuous power consumption of state-of-the-art sensor technologies: Commercially available sensors are not smart enough to identify targets of interest without consuming any power and rely on active electronics to detect and discriminate signal of interest. Therefore\, they consume power continuously to monitor the environment even when there is no relevant data to be detected\, which results in a short battery lifetime limited to very few months. This dissertation presents improvements on a new class of zero-power microsystems that fundamentally break the paradigm\, with zero-power consumption\, until awakened by a specific physical signature. This approach is applied to multiple sensing modalities. In particular\, I have experimentally demonstrated zero-power wireless sensors triggered by different physical and chemical quantities such as: infrared radiation; radio frequency signals; acoustic signals and volatile organic chemicals. The capabilities of the zero-power sensors result in a nearly unlimited duration of operation\, with a groundbreaking impact on the proliferation of the internet of things. \n  \nCommittee: \nProf. Matteo Rinaldi (Advisor)Prof. Marilyn MinusProf. Srinivas TadigadapaProf. Zhenyun Qian
URL:https://ece.northeastern.edu/event/sila-deniz-calisgans-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20221014T150000
DTEND;TZID=America/New_York:20221014T160000
DTSTAMP:20260423T064101
CREATED:20221103T191427Z
LAST-MODIFIED:20221103T191427Z
UID:5932-1665759600-1665763200@ece.northeastern.edu
SUMMARY:Meruyert Assylbekova's PhD Dissertation Defense
DESCRIPTION:“Aluminum Nitride and Scandium-doped Aluminum Nitride materials and devices for beyond 6 GHz communication” \nAbstract: \nWith almost all of the sub-¬6 GHz spectrum now being allocated\, current bandwidth shortage has motivated the exploration of untapped frequencies beyond 6 GHz for future broadband wireless communication. Shift to higher frequency spectra is expected to deliver a significant performance improvement in network capacity\, data rates\, latency\, and coverage. These refinements will enable the development of new life¬changing technologies such as Vehicle to Everything (V2V to V2X)\, ubiquitous Internet of Things (IoT)\, and Augmented and Virtual reality (AR and VR). Among a variety of novel 5G applications\, the implementation of 5G mobile broadband imposes especially demanding specifications on Radio Frequency Front¬End (RFFE) architectures. 5G smartphones are expected to carry over the legacy sub-¬6 GHz bands\, which translates into an increased number of filters. In this context\, the first part of this work will introduce lithographically defined Aluminum Nitride (AlN) piezoelectric microacoustic resonators as a promising solution for the implementation of future minituarized adaptive RFFEs. While AlN has been a material of choice for acoustic filters for over two decades\, future technologies are calling for a material with superior piezoelectric strength. It has been shown that the piezoelectric activity of AlN can be enhanced by partially substituting Al with Sc to form AlScN. Thus\, the second part of this work will explore material properties of AlScN along with the challenges that need to be addressed to take full advantage of its piezoelectric and ferroelectric strength. Last\, AlScN resonators and filters will be demonstrated as promising candidates for the future beyond 6GHz technologies. \nCommittee: \nProf. Matteo Rinaldi (advisor) \nProf. Nicol McGruer \nProf. Cristian Cassella
URL:https://ece.northeastern.edu/event/meruyert-assylbekovas-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220929T090000
DTEND;TZID=America/New_York:20220929T220000
DTSTAMP:20260423T064101
CREATED:20221103T185118Z
LAST-MODIFIED:20221103T185118Z
UID:5930-1664442000-1664488800@ece.northeastern.edu
SUMMARY:Priyangshu Sen's PhD Dissertation Defense
DESCRIPTION:Location: ISEC 532 \n“Physical Layer Design for Ultrabroadband Terahertz Communications: From Theory to Experiments” \nAbstract: \nTerahertz (THz)-band (0.1 THz to 10 THz) communication is envisioned as a key technology to meet the demand for faster\, more ubiquitous wireless communication networks for the sixth generation (6G) of wireless systems and beyond. For many years\, the lack of compact\, fast and efficient ways to generate\, modulate\, detect and demodulate THz signals has limited the feasibility of such communication systems. Recent progress within different device technologies is finally closing the terahertz technology gap and enabling\, for the first time\, experimental wireless research in the THz band. \nThis thesis presents the first steps towards advancing the development and bridging the gap between theoretical and experimental THz communication research. At the core of this work\, the TeraNova platform\, i.e\, the first testbed for ultra-broadband wireless communications at THz frequencies in the world\, is designed and built. In terms of hardware\, the platform consists of multiple sets of analog front-ends at three different frequencies between 100 GHz and 1.05 THz and three different digital signal processing back-ends\, able to manipulate tens of GHz of bandwidth. In terms of software\, tailored framing\, time synchronization\, channel estimation\, and single and multi-carrier modulation techniques are implemented in guided by the experimental characterization of the THz hardware and the THz channel. Moreover\, implementation details and early experimental results to demonstrate the platform’s capabilities/limitations are reported. The platform is then used to demonstrate several milestones in the field\, including the first true THz link in the first absorption-defined window above 1 THz (i.e.\, 1-1.05 THz) and the longest multi-kilometer link (2.01 Km) at the 200-240 GHz band. Further\, Knowing the peculiarities of the THz band and the available device technology in the frequency range\, innovative solutions are proposed. Based on the observed behaviors\, M-ary amplitude and phase-shift keying is presented to simultaneously overcome the limitations due to peak to average power ratio (PAPR) and reduce the effective symbol error rate (SER)\, both while using a high-order modulation scheme. Based on the unique molecular absorption at THz frequencies\, two innovative modulation schemes are presented to make the most of the THz channel. First\, to not only overcome but exploit the distance-dependent bandwidth of the THz band\, hierarchical bandwidth modulations are proposed as a suitable candidate for a single transmitter and multiple receiver (STMR) system. Second\, to reliably communicate even in the presence of absorption peaks\, chirp spread spectrum (CSS) based communication is investigated. Specifically\, Chirp-Spread Binary Phase-Shift Keying (CS-BPSK) is proposed over traditional Binary Chirp Spread Spectrum (BCSS) to obtain better BER. Moreover\, beyond the physics\, current spectrum allocations break down the otherwise very large bands into narrow sub-sets to accommodate sensing users. Spectrum sharing is needed to make the most out of the spectral resources. Therefore\, the capability of the direct sequence spread spectrum (DSSS) is explored to illustrate the performance by acknowledging the coexistence between active and passive users. Further\, channel sounding is conducted in diverse indoor and outdoor scenarios to understand the channel statistics and design reliable communication links. As the last contribution in this dissertation\, the channel model and statistics are explored for an ultra-broadband outdoor channel in different weather conditions. Further\, the channel metrics are explored in various indoor scenarios with different structural and geometrical aspects\, occupancy\, and antenna gain at 130 GHz. For this purpose\, a fully tailored signal processing back-end for sliding correlator type channel sounder is developed\, which is capable of capturing multipath profiles to describe the ultra-broadband nature of the link. \nIn a nutshell\, this dissertation presents the technologies and the results\, highlights the challenges\, and defines a path to move forward with innovative solutions toward practical THz ultra-broadband and long-distance communication systems. \nCommittee: \nProf. Josep Jornet (Advisor) \nProf. Tommaso Melodia \nProf. Milica Stojanovic \nProf. Kaushik Chowdhury
URL:https://ece.northeastern.edu/event/priyangshu-sens-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220926T143000
DTEND;TZID=America/New_York:20220926T153000
DTSTAMP:20260423T064101
CREATED:20221103T184922Z
LAST-MODIFIED:20221103T184922Z
UID:5926-1664202600-1664206200@ece.northeastern.edu
SUMMARY:Michele Pirro's PhD Dissertation Defense
DESCRIPTION:“AlScN material characterization for MEMS applications” \nAbstract: \nThe increasing demand for data is pushing the MEMS industry to more performant and area-efficient systems to be used in IOT nodes as sensors and RF-components. In this market\, AlN plays a pivotal role thanks to the piezoelectric properties accompanied with good stability over power and temperature in miniaturized devices. In fact\, AlN is already present in different commercial MEM systems\, such as duplexers\, ultrasound generators\, energy harvesters and so on\, proving a mature mass-production process flow. The required more stringent specifications in terms of bandwidth\, losses and efficiency are pushing towards piezoelectric materials with higher coupling coefficient\, but still in a compatible post-CMOS process flow. Recent works showed how it is possible to enhance the piezoelectric effect by doping AlN with Scandium\, allowing up to 400% increase in the d33 piezoelectric coefficient. The enhanced acoustic transduction along with the recent demonstration of ferroelectric switching and the post-IC compatibility\, is making Sc-doped AlN a new material with the potential not only to replace AlN\, but also to integrate different functionalities within the same component. Academy and industry all over the world are actively researching the actual potential of the material but there is still a lack of information on high-Sc concentration\, which would allow lower-voltage switching along with higher d33. This work has the main objective to show Sc-concentration > 28% and their piezo/ferroelectric response for a new class of microelectromechanical devices.\nI will discuss the advance in the process flow of high Sc-concentrations\, showing the impact of the deposition parameters on the material properties with focus on the ferroelectric behavior. Effect of RF powers on the substrate and on the target are also analyzed\, demonstrating the possibility to properly optimize the AlScN deposition. Last\, I will present MEM devices which exploit the enhanced piezoelectric activity (high frequency resonators and pmut) and the ferroelectric properties (impedance with memory)\, confirming the potential of the material for new multi-functionalities MEMS. \nCommittee: \nProf. Matteo Rinaldi (Advisor) \nProf. Cristian Cassella \nProf. Siddhartha Ghosh
URL:https://ece.northeastern.edu/event/michele-pirros-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220923T120000
DTEND;TZID=America/New_York:20220923T130000
DTSTAMP:20260423T064101
CREATED:20221103T181851Z
LAST-MODIFIED:20221103T181851Z
UID:5882-1663934400-1663938000@ece.northeastern.edu
SUMMARY:Eric Robinson's PhD Proposal Review
DESCRIPTION:“Techniques for the Modelling\, Design\, and Fabrication of Ultra-Wideband Dipole Arrays”\nAbstract: \nA novel set of techniques are proposed which advance the state of the art for the modelling\, design\, and fabrication of ultra-wideband dipole arrays. First\, existing techniques and relevant topics in the field are introduced and summarized. These include equivalent circuit and Green’s Function models for the impedance of the infinite dipole array. Challenges for the realization of arrays are discussed\, including finite array effects\, common-mode effects\, and the limitations of different fabrication techniques. Several relevant recent innovations by the author are presented to form the foundation of the proposed work. \nFirst\, a new lossy transmission line model for the infinite dipole array impedance is described which results in highly accurate predictions across wide bandwidths and for large scan angles. The accuracy of the model is demonstrated via comparisons to full-wave simulations\, and the model is used to rapidly design a wide-scanning\, ultra-wideband array to demonstrate its value. Next\, a new technique developed by the author is described for physically realizing complex dipole array geometries with integrated dielectrics. A tightly-coupled array is achieved by 3D-printing an array of elements featuring internal through-cavities in the shape of the radiating elements. The internal surfaces of these cavities are then metallized via a copper deposition\, producing an array of conductive elements within a dielectric shell\, resulting in improved mechanical rigidity\, improved scan performance\, and increased inter-element capacitance for ultrawide bandwidth. \nBuilding on these results\, additional research is proposed for completion of the dissertation. First\, the lossy transmission line model will be extended to other relevant cases\, including the unbalanced feed-structure-fed dipole array and the hybrid slot-dipole array. Second\, a new model will be implemented to describe the effects of surface waves on the active impedance of the tightly-coupled dipole array. Mitigating techniques will be proposed based on the results of this model\, enabling the realization of finite arrays which better conform to the predicted performance of the infinite array. Finally\, each of the aforementioned models and techniques will be leveraged towards the design and fabrication of a notional array for an imaging application\, demonstrating their practical value for array design. \nCommittee: \nProf. Carey Rappaport (Advisor) \nProf. Josep Jornet\nProf. Edwin Marengo Fuentes \nDr. Ian McMichael
URL:https://ece.northeastern.edu/event/eric-robinsons-phd-proposal-review/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220922T130000
DTEND;TZID=America/New_York:20220922T140000
DTSTAMP:20260423T064101
CREATED:20221103T184721Z
LAST-MODIFIED:20221103T184721Z
UID:5924-1663851600-1663855200@ece.northeastern.edu
SUMMARY:Justin Crabb's PhD Proposal Review
DESCRIPTION:“Multiphysics Simulation of Graphene Transistors for On-Chip Plasmonic THz Signal Generation and Modulation” \nAbstract: \nTerahertz communication is envisioned as a key technology not only for the next generation of macro-scale networks (e.g.\, 6G+)\, but also for transformative networking applications at the nanoscale (e.g.\, wireless nanosensor networks and wireless networks on chip). This proposal focuses on the development of a multiphysics simulation platform for a plasmonic THz nanogenerator with on-chip modulation. The in-house developed finite-element-method platform\, which self-consistently solves the hydrodynamic and Maxwell’s equations\, is utilized to provide extensive numerical results demonstrating the device’s functionality along with ultra-wide bandwidth and high modulation index capabilities. \nFirst\, a comprehensive theory of the Dyakonov-Shur (DS) plasma instability in current-biased graphene transistors is presented. Using the hydrodynamic approach\, equations describing the DS instability in the two-dimensional electron fluid in graphene at arbitrary values of electron drift velocity are derived. These non-linear equations together with Maxwell’s equations are used for numerical analysis of the spatial and temporal evolution of the graphene electron system after the DS instability is triggered by random current fluctuations. Conditions necessary for the onset of the DS instability and the properties of the final stationary state of the graphene electron system are analyzed. \nNext\, a detailed numerical analysis of the DS plasma instability in the DC current-biased graphene transistor with the gate shifted with respect to the middle of the transistor conducting channel is presented. The geometric asymmetry is shown to be sufficient to trigger the DS instability in the two-dimensional electron gas in the transistor channel. Sustained plasma oscillations in the instability endpoint are demonstrated and the properties of these oscillations are analyzed for different positions of the gate and at different values of other physical and geometric FET parameters. The obtained results show the possibility of designing a tunable on-chip source of THz electromagnetic radiation based on the graphene transistor with a shifted gate. \nFollowing\, the on-chip THz nano-generator with amplitude and frequency modulation capabilities is presented. The proposed device uses and leverages the tunability of the Dyakonov-Shur instability for the growth and modulation of plasmonic oscillations in the two-dimensional electron gas channel of the graphene transistor. \n  \nCommittee: \nProf. Josep Jornet (Advisor) \nProf. Tommaso Melodia \nProf. Matteo Rinaldi \nProf. Hossein Mosallaei
URL:https://ece.northeastern.edu/event/justin-crabbs-phd-proposal-review/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220825T140000
DTEND;TZID=America/New_York:20220825T150000
DTSTAMP:20260423T064101
CREATED:20221103T181955Z
LAST-MODIFIED:20221103T181955Z
UID:5884-1661436000-1661439600@ece.northeastern.edu
SUMMARY:Tarik Kelestemur's PhD Dissertation Defense
DESCRIPTION:Location: ISEC 532 \n“Combining Classical and Learning-based Methods for Visual and Tactile Manipulation” \nAbstract: \nRobots that operate in dynamic and ever-changing environments need to make sense of their surroundings and act in them safely and efficiently. This requires the integration of multiple sensory modalities such as visual and tactile. Humans can naturally fuse different feedbacks from the environment or substitute them with one another to perform everyday tasks. For example\, to use a computer mouse\, we first locate the mouse using vision and then use touch feedback from our fingers to precisely localize the buttons. Ideally\, we would like robots to have human-level perception and control of the environment to achieve various tasks. This dissertation address two significant problems toward this overarching goal. \nThe first problem we consider in this dissertation is figuring out how to use tactile information in conjunction with visual feedback. Robotic manipulators that interact with objects and environments are often equipped with visual sensors such as RGB and depth cameras. They estimate the state of their environment using these sensors and act upon the estimated state. While a large body of previous work has shown that we can achieve impressive results only with visual sensors\, more precise and delicate tasks require touch information which gives direct feedback from the environment. To this end\, we propose methods for efficiently combining the tactile and visual information to leverage the advantages of these modalities.\nThe second problem we investigate is how to build visual and tactile manipulation methods that can generalize over the different novel environments and objects. The rise of deep learning has enabled robots to solve challenging perception and control problems using visual and tactile observations while generalizing to novel objects and environments. However\, a common issue among deep learning-based methods is that these methods usually work only within the distribution of the training data and do not perform well when they are presented with unseen examples. Furthermore\, they cannot distinguish whether they are dealing with in or out-of-distribution data. We propose to address this issue by combining well-established and principled algorithmic priors with the generalization capabilities of deep learning. \nIn the first part of this dissertation\, we investigate the problem of pose estimation of the robotic grippers with respect to the environment and objects. The proposed framework introduces a learnable Bayes filter that can estimate the position of a gripper in a single image of the environment. We learn the observation and motion models of the Bayes filter using modern neural network architectures and use recursive belief updates for tracking the position of the gripper over time. Later\, the belief estimation is used as an input to policies where the aim is to solve manipulation tasks using tactile feedback. In the second part\, we look at the problem of estimating shapes from partial observations. We propose a method called DeepGPIS that combines a powerful deep learning-based implicit shape representation with a non-parametric inference approach model for implicit surfaces (GPIS) which allows us to generate complete shapes of novel objects and estimate their predictive uncertainties. \nCommittee: \nProf. Taskin Padir (Advisor) \nProf. Robert Platt (Advisor) \nProf. David Rosen (Advisor)
URL:https://ece.northeastern.edu/event/tarik-kelestemurs-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220822T150000
DTEND;TZID=America/New_York:20220822T160000
DTSTAMP:20260423T064101
CREATED:20221103T182044Z
LAST-MODIFIED:20221103T182044Z
UID:5886-1661180400-1661184000@ece.northeastern.edu
SUMMARY:Hamed Mohebbi Kalkhoran's PhD Dissertation Defense
DESCRIPTION:“Machine learning approaches for classification of myriad underwater acoustic events over continental-shelf scale regions with passive ocean acoustic waveguide remote sensing” \nAbstract: \nUnderwater acoustic data contain a myriad of sound sources. Among underwater acoustic events\, marine mammal vocalization classification is one of the most challenging problems due to their transient broadband calls\, high variation in the calls of a specie\, and high similarity between the calls of some species. Here\, we developed machine learning approaches for classifying marine mammal vocalizations for real-time applications. We utilize acoustic data from a 160-element coherent hydrophone array and employ the passive ocean acoustic waveguide remote sensing technique to enable sensing and detections over instantaneous wide areas more than 100 km in diameter from the array. A variety of computational accelerating approaches\, combining hardware and software\, that make the methods desirable for real-time applications are also developed. \nThe humpback whale vocalizations can be divided into two classes\, song and non-song calls. Here we use wavelet signal denoising and coherent array processing to enhance the signal-to-noise ratio. To build features vector for every time sequence of the beamformed signals\, we employ Bag of Words approach to time-frequency features. Finally\, we apply Support Vector Machine (SVM)\, Neural Networks\, and Naive Bayes to classify the acoustic data and compare their performances. Best results are obtained using Mel Frequency Cepstrum Coefficient (MFCC) features and SVM which leads to 94% accuracy and 72.73% F1-score for humpback whale song versus non-song vocalization classification. \nTo classify a large variety of whale species vocalizations\, we extracted time-frequency features from Power Spectrogram Density (PSD) of the beamformed signals. Then we used these features to train three classifiers\, which are SVM\, Neural Networks\, and Random forest to classify six whale species: Fin\, Sei\, Blue\, Minke\, Humpback\, and general Odontocetes. We also trained a set of Convolutional Neural Networks (CNN) to detect and classify each of these six whale vocalization categories directly using Per-Channel Energy Normalization (PCEN) spectrograms. Best results were obtained with Random forest classifier\, which achieved 95% accuracy\, and 85% F1 score. To detect transient sound sources\, first we applied PCEN on the PSD of the beamformed signals. We applied thresholding on the PCEN data followed by morphological image opening to find potential sound sources and reduce noisy detections. Then we applied connected component analysis to obtain the final detected sounds for each bearing. To estimate the Direction of Arrival (DoA) of detected sounds\, we applied non-maximum suppression (NMS)\, which is widely used in object detection applications in computer vision\, on the detected sounds. We used mean power of each detected sound as the scores for NMS. To speed up the data processing\, we investigated a variety of accelerating approaches\, such as analyzing the effect of floating point precision\, applying parallel processing\, and implementing fast algorithms to run on GPU. During an experiment in the U.S. Northeast coast on board the US research vessel RV Endeavor in September 2021\, we utilized the software and hardware advances developed here to record underwater acoustic data using Northeastern University in-house fabricated large aperture 160- element coherent hydrophone array with sampling frequency of 100 kHz per element. \nCommittee: \nProf. Purnima Ratilal (Advisor) \nProf. Themistoklis Sapsis \nProf. Devesh Tiwari
URL:https://ece.northeastern.edu/event/hamed-mohebbi-kalkhorans-phd-dissertation-defense/
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20220817T110000
DTEND;TZID=America/New_York:20220817T120000
DTSTAMP:20260423T064101
CREATED:20221103T182252Z
LAST-MODIFIED:20221103T182252Z
UID:5890-1660734000-1660737600@ece.northeastern.edu
SUMMARY:Mithun Diddi's PhD Dissertation Defense
DESCRIPTION:“Multiple UAVs for Synchronous – Shared Tasks and Long-term Autonomy” \nAbstract: \nThis thesis focuses on the use of multiple unmanned aerial vehicles(UAVs) in a distributed framework from a systems perspective to synchronously perform shared tasks such as aerial beamforming and coordinated mapping and to enhance the reliability of performing periodic (mapping) tasks at remote locations for long-term autonomous(LTA) missions. We present an autonomy stack for multiple\, heterogeneous UAVs with a simulation framework. We implemented the end-to-end pipelines for perception and communication applications involving multiple UAVs. \nRepeated deployments in harsh-weather\, real-world locations are challenging and are limited by the need for human operators. These infrastructure-poor\, remote locations pose unique challenges to long-term autonomous missions. In these locations\, harvesting power onboard using solar panels may be a viable alternative for recharging batteries.\nIn the first part of the thesis\, we focus on hardware architecture for UAVs to enable reliable LTA missions with minimal human intervention. We developed a Size\, Weight\, and Power(SWaP) constrained Smart charging stack to minimize hotel loads seen during the recharging process and enable efficient charging of batteries. This leads to the design of a standalone\, solar rechargeable quadcopter.\nReal-world applications such as reconstructing a dynamic scene from multiple viewpoints and distributed aerial beamforming require multiple robots(agents) to coordinate and synchronously act to accomplish shared tasks. These tasks require spatially distant\, multiple UAVs to have time\, phase\, and frequency synchronization. We demonstrate a Synchronous UAV(S-UAV) architecture for wireless synchronization based on GPS disciplined oscillators and the associated software framework needed for temporal registration of data across multiple UAVs.\nWe have built four S-UAVs and demonstrate the ability to 3D reconstruct a dynamic scene from overlapping viewpoints. Dynamic baseline camera arrays formed using multiple S-UAVs are used to synchronously capture a dynamic environment with people moving around. A single-time instance of synchronously captured images of the scene is used to 3D reconstruct the dynamic environment while preserving static scene assumptions of Structure from Motion(SFM). \nIn the second part of the thesis\, we focus on multi UAV autonomy framework for real-world applications of UAVs in perception\, wireless communications\, and reliable LTA missions. We present ‘Simplenav\,’ a navigation stack for heterogeneous\, multiple UAVs\, and ‘OctoRosSim\,’ a computationally lightweight multi-UAV simulation framework for validating the multi-UAV planning and autonomy pipeline. We demonstrate this framework with novel applications of end-to-end autonomy pipelines developed for a coordinated swarm of UAVs. \nCommittee: \nProf. Hanumant Singh (Advisor) \nProf. Kaushik Chowdhury \nProf. Taskin Padir
URL:https://ece.northeastern.edu/event/mithun-diddis-phd-dissertation-defense/
LOCATION:432 ISEC\, 360 Huntington Ave\, Boston\, MA\, 02115\, United States
GEO:42.3396156;-71.0886534
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=432 ISEC 360 Huntington Ave Boston MA 02115 United States;X-APPLE-RADIUS=500;X-TITLE=360 Huntington Ave:geo:-71.0886534,42.3396156
END:VEVENT
END:VCALENDAR